diagnose

Systematic bug diagnosis requires building a deterministic reproduction loop and testing falsifiable hypotheses. This enforces a disciplined 6-phase workflow that eliminates ad-hoc investigation by guiding you from reproduction through hypothesis validation and regression testing.

Root cause analysis for performance regression involves targeted instrumentation and building fast feedback loops. This workflow enforces hypothesis testing and regression tests to validate and lock in fixes, replacing random code inspection with a structured, repeatable debugging process.

Creating a deterministic reproduction loop for flaky failures utilizes ranked strategies like test harnesses, trace replay, and bisection setups. These approaches build fast, agent-runnable feedback loops that reliably trigger intermittent defects for consistent investigation.

Skipping phases in the debugging workflow requires explicit justification. The disciplined 6-phase loop enforces progression from reproduction loop building through cleanup and post-mortem, ensuring no critical diagnostic steps are bypassed without documented reasoning.

Hypothesis testing works for diagnosing hard bugs in high-load server environments by generating falsifiable assumptions and applying targeted instrumentation. This isolates root causes of intermittent lag spikes or failures under load, replacing inefficient random log inspection.

Random code inspection fails to identify non-deterministic defect root causes because it lacks a deterministic reproduction loop. Without structured hypothesis testing and targeted instrumentation, ad-hoc investigation cannot reliably isolate intermittent failures or validate fixes.